A growing body of evidence suggests that sleep facilitates and is beneficial to perceptual learning. However, the underlying mechanism of this facilitatory action is largely unknown. There are two possible types of processing during sleep that may account for the facilitatory action: use-dependent processing and learning- consolidation processing. The use-dependent processing occurs during sleep in the brain mechanisms that are generally used during wakefulness prior to sleep. This processing leads to general changes of neural processing and does not occur specifically for the sake of learning. On the other hand, the learning- consolidation processing works specifically for learning. It is highly controversial concerning whether the use- dependent processing is sufficient for the facilitatory action or whether learning-consolidation processing is necessary for the facilitatory action. It is fundamentally important to know which type of processing occurs during sleep to clarify the mechanism of sleep facilitating perceptual learning, since it has not been directly tested which model is valid. We address this question in the present proposal. Sleep consists of different dynamics, such as those reflected by a multitude of frequency bands in spontaneous brain oscillations in each rapid eye movement (REM) and non-REM (NREM) sleep. This raises the possibility that only one type of processing does not necessarily occur consistently throughout the whole sleep period: the learning-consolidation processing might occur for some frequency bands in some cortical areas, while use-dependent processing might occur for others. Thus, we will systematically examine whether learning-consolidation processing or use-dependent processing occurs in each band in each REM and NREM sleep, and in different brain area(s). To test whether the learning-consolidation processing is necessary for facilitating perceptual learning during sleep, we will compare the spatio-temporal brain activation patterns during sleep that follows task performance that causes learning (learning paradigm) with those during sleep that follows task performance that does not cause learning (interference paradigm). For this purpose, we must obtain highly localized spatio- temporal information about brain activation during sleep; we will use a cutting-edge neuroimaging technique that combines fine temporal information from magnetoencephalography (MEG) and electroencephalography (EEG) with fine spatial information from magnetic resonance imaging (MRI) as well as individual retinotopic mapping in the early visual areas, to estimate the power and phase information of spontaneous oscillatory activities in the precisely localized cortical regions. Imaging will be conducted with concurrent polysomnography measurement to objectively identify sleep stages. Successful research results would provide significant knowledge to clarify how improvement of perceptual learning of a visual task and visual plasticity occurs during sleep after training of the task.